Website outage

I discovered a problem with my blog when I posted my rant about my Cisco RV016 earlier today. After all my network shuffling around, the website was loading very, very slowly. The WordPress blog was taking several minutes to render, whereas the other blogs on the server were behaving normally.

I tried a great many things, including running database recovery diagnostics and upgrading my WordPress version, before finally figuring out the cause. My network changes resulting in my server no longer having correct DNS settings: you could find it, but it couldn’t find the rest of the internet. I suspect the page generation delay was happening because of the Google Adsense stuff on the site, which wants to talk to Google… but I’m just speculating.

In any case, the problem appears to be solved now… five hours after I discovered it. I’m going to add this to the list of things I will blame on my Cisco RV016 router/switch…

UPDATE: I have uncovered and repaired more problems, but these ones I can’t blame on my much maligned Cisco RV016

  • WordPress kept failing to generate pages, particularly when I went to the Admin page. I found that the mySQL database tables that support WordPress had become corrupted by looking at the PHP log files and finding an error like this:
  • Apr 5 17:43:11 kgadams httpd: WordPress database error Table ‘./wordpress/wp_comments’ is marked as crashed and should be repaired for query SELECT COUNT(*) FROM wp_comments WHERE comment_approved = ‘0’ made by require

    I suspect this was the result of a power brownout/server crash on Friday. I used the standard mySQL database recovery techniques, but it took a few tries going through their sequential “try this first… then try this” process. The last table I had to recover, the wp-comments table, required a “stage two” repair with the –safe-recover option

  • After completing the mySQL table recovery actions, Apache started generating core dumps (and WordPress continued to have problems generating pages). The relevant error appeared this time in the HTTP log files (error_log):
  • [Sat Apr 05 18:02:26 2008] [notice] child pid 5423 exit signal Segmentation fault (11), possible coredump in /tmp

  • The new version of WordPress (2.5) required a small tweak to its default memory settings- I updated the value of WP_MEMORY_LIMIT in wp-settings.php from 32M to 64M. Note that I have a fair number of plugins running

It appears that making these corrections has eliminated the problems. I hope…

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.